Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
Digital Twin for a Vehicle: ElectroBus Case Study
Diego M. Botín-Sanabria, Diego A. Santiesteban-Pozas, Guillermo Sáenz-González,
Ricardo A. Ramírez-Mendoza, Mauricio A. Ramírez-Moreno, Jorge de J. Lozoya-Santos
School of Engineering and Science, Mechatronics Engineering Department
Tecnológico de Monterrey, Campus Monterrey
Eugenio Garza Sada 2501 Sur, Monterrey, Mexico
botin@tec.mx , mauricio.ramirezm@tec.mx , jorge.lozoya@tec.mx
Abstract
A Digital Twin is a virtual representation of a real dynamic system that can simulate its current conditions, predict its
future behavior, and log valuable information about its internal operation and interactions with other systems. The
unique capability of automatic bidirectional information flow between virtual and physical worlds and predictive
analysis of Digital Twins are what make this emergent technology greatly innovative and of significant value. The
objective of this case study is to implement a vehicle’s performance Digital Twin concept on an electric passenger
bus and analyze the technology’s adaptability level, capabilities, scope, limits, and future improvements. This system
incorporates a network of sensing devices mounted on the vehicle, a real-time simulation model, edge computing
capabilities and the possibility of incorporating predictive maintenance models to determine remaining useful life of
components. The Digital Twin is thought to have great value when it comes to gaining deep insight on the dynamic
performance of the bus, analyzing energy waste, exploring determining factors of energy usage and monitoring the
current and future state of the vehicle. All of this being possible through simulation technology, digital modeling,
physical and virtual sensors, edge computing, Internet-of-Things networks, and Machine Learning algorithms.
Keywords
Digital twin, simulation, vehicle dynamics, case study, internet of things.
1. Introduction
Digital twins (DT) are an emergent technology which has seen a recent surge in use cases and applications in a variety
of industries. Due to the benefits of developing DTs for specific applications, it a desirable method for gaining deeper
insight and testing/validation of physical object or processes. By having a reliable and experimentable model as a
virtual testbed, users can extract information about current inner operations of a real object or process, predicting its
future behavior and even simulating what-if situations. In this sense, it is possible to develop high fidelity,
comprehensive “Simulation Models allowing a simulation-based verification and validation throughout the whole
lifecycle (of the physical twin)” (Dahmen and Rossman, 2018). Recent studies and use cases for DTs revolve in its
majority on manufacturing and smart city applications. However, with the emergent trend of electrification and
evermore complex mechatronic systems, there is an opportunity of developing DT concepts for automotive
applications. This work presents the case study of the use of the DT concept of a vehicle applied on an electric bus
(Electrobus). This bus was converted from and internal combustion engine (ICE) to a fully electric (EV) powertrain
and is expected to become a vehicle with smart mobility and connectedness capabilities. Amongst these capabilities,
is the DT concept of the bus which would serve as a way of monitoring its performance, having an experimentable
model of the physical bus, being able to predict future performance behavior and making available for the driver and
other users, deep insight on sustainability and performance analysis of the vehicle and its interaction with its
surroundings. The DT of a vehicle was previously developed for the representation of a commercial ICE vehicle. The
original test vehicle was the Mazda 3 2018 hatchback which is classified as a Class C mid-range road family car. For
this work however, the DT platform was converted and adapted to run the simulation model of an electric bus and the
focus of the data analysis was shifted towards sustainable mobility. The test vehicle for this work is a Class 6 (gross
vehicle mass of approximately 10,000 kg) city bus modified and converted to an EV bus. The powertrain, body work
and driver cab will be modified, but the chassis for the bus will remain the same. The result will be the capability of
analyzing the performance of the bus in real-time and having a record of its performance through time. This will
enable deeper insight of the inner behavior of the vehicle. This in turn will provide information for data-led decision
© IEOM Society International 2971
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
making in terms of maintenance, driving styles and the understanding of the energy consumption and environmental
impact of an EV mass transportation service in Monterrey.
The DT is intended to be optimized for studies on the use of energy by the bus under different conditions, the dynamic
performance of the bus and the impact it has on its surroundings. In this sense, working under the scope of the United
Nation’s (UN) Sustainable Development Goals (SDGs) was important to determine the objectives of the system and
the focus of the DT. More on the SDGs in Section 1.1.
This research is of great value since it is one of a few publications made on performance DTs for vehicle models that,
furthermore, analyzes the results and behavior of the physical bus under a sustainability and smart mobility framework.
Additionally, the DT of the vehicle is intended to be able to generate, populate, and interact with the DT of an urban
space. This concept of interacting DTs will demonstrate the scope of this technology and open possibilities for new
implementations and developments in the subject.
1.1 Objectives
DT systems are complex and often hard to implement due to a variety of limitations and challenges. Amongst, these
challenges, the most common are costs related to implementation (hardware), interoperability of platforms and
devices, and a great investment when it comes to staying up to date with the development of enabling technologies
such as IoT, ML and edge computing. The developed system for a performance and urban space DT is expected to be
put to test in a real situation where these challenges and limitations can be documented, explored, and targeted in the
most efficient way. The vehicle is expected to generate and populate the urban space DT using IoT sensing devices,
edge processing and modeling and simulation software. The main objective of this work is to demonstrate the
implementation of a DT for a vehicle and its interaction with the generated DT of an urban space. This type of DT
interaction demonstrations is of great value to developing the DT concept and proposing the methodology for their
design and development. This first step of the project will provide a framework and baseline to work with this type of
technology which effectively contributes to the standardization and widespread design of DTs.
For the design and implementation of this DT concept under the framework of sustainability, UN’s SDGs were
consulted. This work’s objectives have been aligned with some global issues as presented by the following goals:
• Goal 3 – To ensure healthy lives and promote well-being for all at all ages.
o Through the implementation of urban and vehicle DTs, one can perform studies and analysis of an urban
space in terms of security, infrastructure, accessibility, open spaces, etc. to determine living standards and
community perceptions.
• Goal 9 – To build resilient infrastructure, promote inclusive and sustainable industrialization and foster
innovation.
o The digital representation of an urban space may showcase the evolution of a community and effectively
gather historical information on variables such that a deeper insight of the space is available. This level of insight
may improve city planning, logistics and help understand city perceptions.
• Goal 11 – To make cities and humans settlement inclusive, safe, resilient, and sustainable.
o According to the UN, “as of 2020, 16% of the average global share of urban areas was allocated to streets
and open public spaces. This is short of the 30% streets and 10-15% public open spaces target” and “as of 2019,
only half of the world’s urban population have convenient access to public transport. Convenient access means
residing within 500 m walking distance of a bus stop/ low-capacity transport system and 1000 m of a railway or
ferry terminal” (Department of Economic and Social Affairs, 2021).
2. Literature Review
Literature reviews are an important step when it comes to staying up to date with the most recent advances in
technology and other subjects. It is an unbiased approach to evaluating the state of the art (SoA) of a certain technology
with evidence of recent applications, implementations, research, and patents. There exists a variety of methodologies
to performing and writing literature reviews, however that which aligns with a systematic approach to such research
is desired to eliminate any kind of bias that might exist. A systematic-like review (SLR) is one that aims to “collate
evidence that fits pre-specified eligibility criteria to answer a specific research question. They aim to minimize bias
by using explicit, systematic methods documented in advance with a protocol" (Higgins and Thomas, 2021). These
authors present a methodology for SLRs that include the following steps.
© IEOM Society International 2972
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
1. Identifying answerable questions
2. Developing a protocol
3. Conducting systematic publication searches
4. Selecting studies to include
5. A comprehensive revision
6. Extracting and synthesizing information
7. Writing and publishing the review
In accordance with the outline methodology, the following questions were determined for this SLR. RQ represent the
research question and SQx represents the sub questions.
• RQ: What is the SoA of DT technology in implementation applications?
• SQ1: What are the current challenges of implementing a DT-based system with current technology?
• SQ2: What are the enabling technologies and their use trends for DTs?
Figure 1 presents the protocol, systematic publication search and selection criteria for the studies to include. For the
systematic search, databases such as MDPI, Research Gate, IEEE Xplore, Science Direct and ProQuest were used for
selecting an initial set of 110 publications that were then collated using the selection criteria. Afterwards, the remaining
set of publications were revised.
Figure 1. Protocol and selection criteria for SLR.
It was found that there currently exists no well defined or universally accepted definition for DTs. According to
Sharma et al. (2020), “this lack of standards impedes the widespread design, implementation and adoption of this
technology”. In this sense, this work aims to contribute to the standardization of the DT concept and definition. It is a
challenging task due to the wide variety of applications and use cases for this technology, but through more thorough
reviews and publications, a well-defined definition is possible. Evans et al. present a maturity spectrum index for the
evaluation of DT application maturity in terms of its integration and connectivity capabilities. Table 1 presents this
spectrum index.
Table 1. DT application maturity spectrum (Evans et al.)
Maturity element Defining principle Urban space outline usage
(logarithmic scale of
complexity and
connectedness)
0 Reality capture (LIDAR, drones, Existing as-built digitalization
photogrammetry, plans, etc.)
1 2D maps/system or 3D model Space coordination and model
© IEOM Society International 2973
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
2 Connect model to persistent (static) data, Asset management, life cycle monitoring,
metadata and BIM stage 2 simulation experimentation
3 Enrich with real-time data Real-time life cycle monitoring
4 Two-way data integration and interaction Remote and immersive operations,
control physical from the digital
5 Autonomous operations and maintenance Complete self-governance with total
oversight and transparency
This work aims to reach the level 3 maturity level where real-time dynamic information is used to enrich the simulation
and models. However, there exists certain restriction and limitations to the widespread adoption and implementation
of DTs. Parrott and Warshaw (2017) argue that “many companies found that the connectivity, computing, data storage,
and bandwidth required to process massive volumes of data involved in creating digital twins were cost-prohibitive”.
Other limitations include the complex task of modelling evermore complex mechatronics systems and staying up to
date with enabling technology advancements such as Machine Learning, Big Data, IoT, and communication systems.
Deloitte Tech Trends (2020) present the fact that developments in simulation and modeling tools, IoT device
conectivity, expanded bandwidth and better computing architectures will enable DTs to become a predominant tool
for companies and governments.
In terms of the classification of DTs, Juarez et al. (2021) propose three classes regarding integration levels:
• Digital Model: virtual representation of a physical subject with no automated flow of information from
physical to virtual worlds.
• Digital Shadow: Virtual representation where there is a unidirectional flow of information usually from
physical to virtual world.
• Digital Twin: Uses a bidirectional flow of information scheme to enable the management and monitoring of
the object’s life cycle.
Singh, et al. (2021) propose a three-level hierarchy classification: the basic level is the Unit Level where DTs represent
single objects, materials, etc. In the System Level, physical twin could be a production process or a complex object’s
lifecycle and finally a System-of-systems (SoS) level where there is product of process life cycle management.
Some recent applications of DTs for the automotive industry include estimating the battery state of electric golf
vehicles (Merkle et al., 2021), a study of DT potential applications in EV technologies (Van Mierlo et al. 2021), DT
concepts for electric powertrain applications (Rodríguez et al., 2021), automotive validation methodologies using DTs
(Szalay, 2021), an even the simulation and monitoring of vehicle sensors (Tavakolibasti et al., 2021).
3. Methods
DTs as presented by Campos-Ferreira et al. (2019) are made up of three main components: a physical world, a digital
world, and the connectivity that allow bidirectional flow of information amongst both worlds. In this sense, the DT
for the Electrobus is designed following this concept where the physical world is represented by the physical vehicle,
the digital world integrates the digital model, simulation and visualization platforms, and the connectivity component
is enabled by the communication capabilities amongst physical and virtual world. The overview of this work’s vehicle
DT is presented in Figure 2. More detail on the devices and methods for each component are listed on Sections 3.1-
3.3.
© IEOM Society International 2974
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
Figure 2: Architecture for vehicle DT.
3.1 Data Acquisition
To collect the necessary data to successfully run the digital model of the vehicle, a wide variety of methods and sensing
devices were used. For instance, there is a need of static data represented by the technical specifications that
characterize the vehicle and dynamic data which is given by a set of sensors and device that gather information in
real-time. For the technical specifications of the vehicle, various parameters were used to define the different systems
and dimensions of the model. For example, information about the body dimensions, nominal values for the motor,
aerodynamic performance of the chassis, and tire specifications. Some of the most valuable information is given in
the Table 2.
Table 2. Electrobus vehicle technical specifications.
Category Specification Value Units Comments
General dimensions Length, width,
[10.34, 2.36, 3.43] [m, m, m]
height
Nominal vehicle Minimum
weight 10,000.00 [kg] functional systems
only
Body
Expected max With passengers
15,000.00 [kg]
weight and extra systems.
Wheelbase 4.75 [m] -
Expected number of
44 - -
seats
Tires Type 275/80 R 22.5 - -
Motor max power
356.20 [kW] -
rating
Drive train
Battery power
500.00 [kW] -
rating
For the dynamic data aspect of the data acquisition network, a set of sensing and ranging device connected to an edge
processor and mounted directly on the vehicle itself are used. A complete list of the devices and their use if presented
in Section 4.
3.2 System Modeling
For the vehicle model and simulation, the software Matlab and Simulink are currently being used. These platforms
present great interoperability with other software and it allows to create a tailored-made model and simulation
environment for our work. It’s capability of hardware deployment with NVIDIA processor through Compute Unified
© IEOM Society International 2975
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
Device Architecture (CUDA) (MathWorks Inc, 2021) is also a desirable aspect. This way, all models and script from
Matlab/Simulink may run on the edge processor.
For the vehicle model itself, a high-fidelity model is expected since there is a need to represent the physical vehicle’s
behavior and inner operations as accurately and precisely as possible. There are many approaches to vehicle and
system modelling; however, for this work, a multisystem, parametric, and data-driven model was designed. This way,
the model is connected to external input from real sensors for the data-driven aspect and is parametrized using the
technical specifications and technical approximations. According to Aivaliotic et al. (2019), there are three modelling
levels: “black (without any knowledge of the internal operation), grey (theoretical data are used to complete the
model), and white boxes (fully described component)”. In this sense, the model is designed mostly in grey and black
box levels. The most crucial systems were modelled in such a way that most parameters are editable.
3.3 Connectivity
To enable connectivity between worlds, an network of IoT devices was design and implemented to allow each to
communicate with each other and with the digital world. Although the data acquisition network and the digital world
are connected through physical cabling, connectivity of the edge processor to an external web server is performed
through wireless data streaming and communication. In this sense, various communication protocols and methods are
used such as serial communication, Ethernet, application programming interfaces (APIs), etc. Furthermore, the
external web server is used to live stream information from the vehicle, display graphic results to an end user and
serves as a database storage system. More on the wireless communication and web server design is explained on
Section 5.
4. Data Acquisition
For the data acquisition network, a total of 5 devices and an edge processor were used. All the sensing devices are
mounted directly on the vehicle itself to reduce the amount of data being transmitted wirelessly and in real-time since
this presents a challenge with current communication technology. Table 3 details the use and data gathered from each
device.
Table 3. Devices used in data acquisition network
Device Use Data acquired
COWTECH device Real-time information directly from the Vehicle speed, engine RPM, battery
car’s controller area networks (CAN) bus. state of charge (SoC), steering
wheel position, throttle/brake
commands, tire pressures.
Inertial Measurement Unit Real-time information from car’s dynamic Accelerometer, gyroscope,
(IMU) behavior. magnetometer.
Light Detection and Offline 3D mapping of surrounding area and Point cloud data.
Ranging (LIDAR) sensor trajectory estimation.
Geotab Device Real-time geolocalization. GPS coordinates.
Camera Real-time object detection using Machine Video frames.
Learning (ML) algorithm.
For this work’s system and architecture, the LIDAR, IMU and camera are used to gather information for the creation
of the digital model of an urban space and effectively enable its DT. Specifically, the LIDAR is meant to be used for
digitalization of spaces in terms of dimensions and objects. The IMU is meant to be fused with LIDAR information
to produce an accurate and precise vehicle trajectory estimation that may be used for deeper mobility and infrastructure
studies. The camera is used to detect objects throughout the vehicle’s trajectory. This information, paired with GPS
information on the detected object may generate valuable information for mobility infrastructure and civilian density
studies.
Lastly, information from the devices reaches the central edge process which is a Jetson Xavier Developer board from
NVIDIA. This processor is built to run Linux and have deployed applications from software like Matlab, Simulink,
Python code, etc. It is also optimized to run Machine Learning vision and navigation algorithms in the most efficient
way. All the devices are meant to be mounted on the vehicle. The specific location for each device may vary; however,
© IEOM Society International 2976
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
it is important to have the LIDAR and IMU devices mounted as closely as possible to the vehicle’s center of gravity
to avoid any bias or drift form their measurements. Furthermore, the camera needs to be mounted on the front part of
the vehicle where its field of view (FoV) is not blocked.
5. Results and Discussion
The proposed system is still under development; however, important advances on the development and
implementation of some systems is already taking place. For instance, the mapping system is currently being tested to
provide insight on its accuracy and precision. The digital model and its connection to the physical vehicle is also being
developed, however, some advances on the modelling of the Electrobus with the associated companies is being made
on a steady pace. Some other options for software and modeling techniques are also being explored but more
information is provided on Section 5. Some current results presented in this section include the web server design, the
mapping system design and experimentation advances as well as the IoT architecture development.
After having tuned the parameters of the LIDAR mapping module, an experimentation process was used to test the
system under different conditions to evaluate and validate its performance. In this sense, an experimentation scenario
was generated where the objective is to test the effect that the vehicle’s speed has on the precision and accuracy of the
2D map and the estimated trajectory. Also, the effect of surrounding dynamic objects (moving vehicles, people,
animals, etc.) is to be evaluated. For this scenario, a straight street (Avenida Fundadores) with little elevation was
chosen. The experiment consists of driving at a constant speed three times. Then changing the speed (20 km/h, 40
km/h and 0-60-0 km/h) and evaluating if there is a difference in percent error with respect to the Google Maps
trajectory distance estimation. Comparison of distance estimations is performed for each set of takes at the same speed
and estimations at different speeds effectively evaluating both precision and accuracy. The results and evaluation of
this experiment are shown in Section 5.1, Table 4.
To achieve more accurate error values, a stricter control of the travelled distance is necessary. For instance, ensuring
that every take of the experiment consists of the same real distance. This will be achieved by using a Garmin device
to calculate real distance and altitude and having start/stop marks on the road to ensure every take starts and ends in
the same place. Also, additional testing is required to evaluate the LIDAR-IMU fusion algorithm.
5.1 Numerical Results
The following table present the results of the LIDAR mapping module experimentation. In the row “Distance”, the
estimated trajectory distance by the system is presented. Table 4 shows that the average percent error for distance
estimation is 1.80% (translates to an average error less than 5 m) which as a positive result that shows a level of high
definition of the mapping module. The expectation for the experimentation of the LIDAR-IMU fusion algorithm is
that this error will be reduced to approximately 1.00% (less than 2.75 m average error). Additionally, GPS coordinates
may be added to the system in the future to further decrease the error and enable other functionalities to the system
such as live GPS tracking of vehicles within the urban space DT (effectively achieving interaction between both DTs).
Table 4. LIDAR mapping module experimentation results.
40 km/h 20 km/h 0-60 Total
km/h
Category 1 2 3 Avg. 1 2 3 Avg. 1 Avg. Variance Standard
Deviation
Distance 315,73 286,08 248,95 283,59 237,17 264,43 254,51 252,04 260,21 266,73 594,64 24,39
[m]
Rise [m] 37,50 36,54 35,00 36,35 35,33 43,87 32,45 37,22 31,32 36,00 14,35 3,79
Predicted 322,00 290,55 253,79 239,39 270,53 259,06 266,12
Dist.
(Google
Maps)
% Error 1,95 1,54 1,91 1,80 0,93 2,25 1,76 1,65 2,22 1,79
RMSE 5,08
Apart from the percent error and RMSE values, this initial experiment was significant in the sense that it demonstrated
that dynamic objects have very little to no effect on the precision of the system. This is mostly because the near
© IEOM Society International 2977
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
distribution transform (NDT) distinctive feature registration (DFR) algorithm only register distinctive and static
features throughout LIDAR frame readings. This means, that the algorithm ignores dynamic object such as cars and
people. With these results, one can also conclude that speed and acceleration have little effect on the precision of the
system. This is more evident with the 0-60 km/h take which shows and increase of percent error. The expectation is
that the LIDAR-IMU fusion algorithm will help reduce this effect attributed to fast movements of the sensor. This due
to the fact the IMU gyroscope might reduce the drift effect in LIDAR data.
5.2 Graphical Results
After processing LIDAR data, the accumulated PC map is generated. Figure 3 portrays the result of one of the Scenario
1 data takes (left) and the Google Maps distance estimation (right). As mentioned earlier, for future experiments and
more accurate results, a Garmin device will be used to set true values of distance and altitude. The result from the
current mapping module is a 2D map and a 3D trajectory estimation, however, after being processed with ArcGIS, the
system will yield HD 3D visualizations of urban spaces and more accurate vehicle trajectory estimations.
Figure 3: LIDAR mapping module result from an experiment in Avenida Fundadores (1.8% traveled distance error).
Apart from the experiment, the system was also tested with single takes (no experimental control) on more complex
trajectories. Figure 4 is an example of a take where the vehicle travelled through a neighborhood with more elevation
differences, speed bumps, sharp turns, and some dynamic objects (people).
Figure 4: LIDAR mapping module result from take inside a neighborhood (Los Angeles, Monterrey).
© IEOM Society International 2978
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
5.4 Validation
The validation process is very important when it comes to evaluating and validating systems or models that represent
real objects or processes. In our case, since the digital models are desired to produce calculations and conclusions on
the behavior of their physical twins, these need to be as accurate and precise as possible to present the most real
information as possible. For our system experiments, the validation process is carried out by comparing absolute true
values to those output by our design system. In the case of the mapping module, distances and altitude is evaluated
through percent error and RMSE calculations.
When the system and individual models are complete, the use of other validation techniques are necessary. For our
work, the Grieve’s Tests of Virtuality (GTV) will be performed to evaluate the ability of our models to mirror their
physical twins. These GTV consists of three tests: sensory visual, performance, and reflectivity (Juarez et al. 2021)
where the behavior similarity of both physical and virtual worlds is evaluated.
• Sensory Visual Test: A tester subject demands a movement either from the DT or the physical twin and if
the tester cannot differentiate among the real system and the DT, this test is approved.
• Performance Test: The tester demands an action from both twins and if it cannot differentiate the performance
among both, this test is approved.
• Reflectivity Test: The tester demands information on the current state of the object, if there is no difference
between the data from the physical and virtual twin, this test is approved
6. Conclusion
The results and designed software for this work already has great value in terms of setting a baseline for the future
development of the DT for a vehicle and for urban spaces. Implementation is still pending, however after the proposed
validation techniques, system integration will be feasible, and implementation would be the next step. In this sense,
the objective for the project is still to be reached. Some challenges and limitations have already been worked around.
For instance, implementation costs due to the number of devices and software needed is a challenge to overcome. For
this, open-source software and popular platforms were used and tailored-made designs developed. Other challenges
like the complexity of interoperability amongst software was experimented. Matlab and Simulink were a solution
option due to their facility to interact with other platforms.
With the current results, the system may already be classified in a maturity level 1 according to the DT maturity
spectrum index. However, once the system is complete and once it is validated, a level 2 and even level 3 of maturity
will be achieved. The static information and real-time information will come from the sensing devices mounted on
the vehicle. This will also demonstrate the interaction between an urban space DT and a vehicle DT concept.
Furthermore, the current system may be classified as a SoS level hierarchy and an integration level of Digital Shadow.
When the system is complete, the integration will be classified as a Digital Twin. A lot of work needs to be done with
the interoperability and connectivity of platforms and devices. It is a special challenge when it comes to real-time data
processing, and enabling technologies such as ML, Big Data, IoT and edge processing have a great role.
Future work will focus on connecting both physical and virtual world and performing experimentation to evaluate and
validate individual systems. After this, implementation of the DT for Electrobus will be enabled.
7. Acknowledgements
For the development and implementation efforts, a total of 5 associated companies are involved. Tecnológico de
Monterrey and this work’s authors are currently collaborating with an automotive bus manufacturer, a bodywork
manufacturer, a public transport agency, and two EV conversion companies.
References
Dahman, U. and Rossman, J., Experimentable Digital Twins for Modeling and Simulation-based Engineering
Approach, IEEE International Systems Engineering Symposium (ISSE), 2018.
United Nations General Assembly, Transforming our World: The 2030 Agenda for Sustainable Development, United
Nations, res. 70, no. 01, pp.14, 2015.
Sharma, M. and George, J., Digital Twin in the Automotive Industry: Driving Physical-Digital Convergence, TATA
Consultancy Services, White Paper, 2018.
Parrott, A. and Warshaw, L. Industry 4.0 and the digital twin, Deloitte Insights, Article, 2019.
© IEOM Society International 2979
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
Deloitte Insights, Digital twins: Bridging the physical and digital, Tech Trends 2020, 2020.
Merkle, L., Pothig, M. and Schmid, F., Estimate e-Golf Battery State Using Diagnostic Data and a Digital Twin,
Batteries, vol. 7, 2021.
Mierlo, J., Berecibar, M., Baghdadi, M., Cauwer, C., Messagie, M., Coosemans, T., Jacobs, V. and Hegazy, O.,
Beyond the State of the Art of Electric Vehicles: A Fact-Based Paper of the Current and Prospective Electric
Vehicle Technologies, World Electric Vehicle Journal, vol. 12, 2021.
Rodríguez, B., Sanjurjo, E., Tranchero, M., Romano, C. and González, F., Thermal Parameter and State Estimation
for Digital Twins of E-Powertrain Components, IEEE Access, 2021.
Szalay, Z. Next Generation X-in-the-Loop Validation Methodology for Automated Vehicle Systems, IEEE Access,
2021.
Tavakolibasti, M., Meszmer, P., Bottger, G., Kettelgerdes, M., Elger, G., Erdogan, G., Seshaditya, A. and Wunderle,
B., Thermo-mechanical-optical coupling within a digital twin development for automotive LiDAR, 22nd
Internation Conference on Thermanl, Mechanical and Multi-Physics Simulation and Experiments in
Microelectornics and Micro-Systems, 2021.
Juarez, M., Botti, V. and Giret, A., Digital Twins: Review and Challenges, Journal of Computing and Information
Science and Engineering, vol. 21, 2021.
Singh, M., Fuenmayor, E., Hinchy, E., Qiao, Y., Murray, N. and Devine, D., Digital Twin: Origin to Future, Applied
System Innovation, vol. 4, no. 36, 2021.
Campos-Ferreira, A.E., Lozoya-Santos, J.J., Vargas-Martinez, A., Ramírez-Mendoza, R. and Morales-Menéndez, R.,
Digital Twin Applications: A review, Memorias del Congreso Nacional de Control Automático, 2019.
MathWorks Inc., GPU Computing, Matlab Services, website: https://www.mathworks.com/solutions/gpu-
computing.html, 2021.
Aivaliotis, P., Georgoulias, K., Arkouli, Z. and Makris, S., Methodology for enabling digital twin using advanced
physics-based modellin in predictive maintenance, 52nd CIRP Conference on Manufacturing Systems (CMS), vol.
81, pp. 417-422, 2019.
Biographies
Diego M. Botín-Sanabria is a Mechatronics Engineering bachelor student in Tecnologico de Monterrey. He is Project
Manager Jr. for the Digital Twins and Digital Cab research programs and technical lead for this work. Diego has
current collaboration with the University of Technology Sydney and Macquerie University, Australia. Diego is the
Chief Engineer for his university’s scuderia: Tec Racing and has experience in C, MATLAB/Simulink, SolidWorks
and CarSim. His research interests are systems modeling and simulation, vehicle dynamics, digital twins, and
motorsports engineering.
Diego A. Santiesteban-Pozas is a bachelor’s student of Mechatronics Engineering at the Instituto Tecnologico de
Monterrey, in Nuevo León, Mexico. He participated in an exchange program at the Technical University of
Brunswick, in Lower Saxony, Germany. His interests include mechatronic design, electronics, automation, Industry
4.0/IoT, and cyber-physical systems.
Guillermo Sáenz-González is a bachelor’s student of Mechatronics Engineering at ITESM, in Nuevo León, México.
Guillermo has participated in a research project with Siemens’ facility in Santa Catarina, Nuevo León. His interests
include programming, electronics, control systems, automation, data analytics, and Industry 4.0.
Ricardo A. Ramírez-Mendoza received his Ph.D. degree in Automation and Production from Grenoble Institute of
Technology, France (1997). He has published over 500 papers in journals, conferences, etc. and has mentored over 40
graduate students, who occupy leading positions in academia & industry. His research lines include Automotive
Control, Active Control, Vehicle Dynamic Control, Mechanical Vibrations, Brain-Computer-Interface, and
Biomedical Signal Analysis. He is Dean of Research and Graduate Studies at the School of Engineering and Sciences,
and Professor of Mechatronics and Mechanical Engineering in Tecnologico de Monterrey.
Mauricio A. Ramírez-Moreno received his PhD in Biomedical Engineering in 2019 in Cinvestav Monterrey
(Mexico). In 2019, he joined the School of Engineering and Sciences at Tecnologico de Monterrey. His main research
interests include Brain-Computer Interfaces, neuroengineering, robotics, biomechanics, smart cities and machine
learning; and has published five journal papers, three conference papers, and one book chapter in the fields of smart
cities and neuroengineering. He is currently a postdoctoral researcher, and the Program Manager of Campus City
Smart Mobility and the IUCRC BRAIN TEC initiatives.
© IEOM Society International 2980
Proceedings of the International Conference on Industrial Engineering and Operations Management
Monterrey, Mexico, November 3-5, 2021
Jorge de J. Lozoya-Santos received his PhD Degree in Mechatronics and Advanced Materials (2013) from
Tecnologico de Monterrey. Jorge has current collaborations with Politecnico di Milano, Italy; Institute Polytechnique
de Grenoble, France; Universita degli Studi di Modena and Reggio Emilia, Italia among others. He has more than 25
international conferences, 10 indexed journals and 4 patent applications. His research interests are intelligent
transportation systems, modeling, and control systems, applied automatic control and automotive systems. He is
Research Professor in Tecnologico de Monterrey, School of Engineering and Sciences.
© IEOM Society International 2981
